Recent technical advances in digital still cameras, digital video cameras and other such image input devices have tended to provide sensors with more and more picture elements (hereinafter pixels) in order to further improve the quality of the images formed, with the result that higher readout speeds are required as well. In order to meet such a need, a readout method has been developed by which the pixel signals have been divided into a plurality of readout channels. A description of this conventional method is now given with reference to FIGS. 11 and 12.
FIG. 11 is a diagram of the structure of a conventional image sensing apparatus. FIG. 12 is a timing chart showing the drive timing and output signals of a conventional image sensing apparatus.
The image sensing apparatus shown in FIG. 11 has a plurality of pixels 3101 arranged in two dimensions and consisting of optical black pixels arranged and shielded by a light-blocking film and effective pixels with no light-blocking film; readout channels 3071 and 3072 for reading out image signals from each of the plurality of pixels selected according to a control signal from a vertical scan circuit 3102; and an output terminal 3120 for outputting signals after increasing either the waveform or the drive force of the signals after the timing with which the signals are read out by the readout channels 3071, 3072 has been adjusted and the signals have been passed through a buffer circuit 3119. Note that only 5×4 pixels are shown in FIG. 11 for simple explanation.
In addition, the readout channels 3071, 3072 have readout circuits 3106, 3111, which in turn have line memories 3104, 3109 for storing pixel signals read from each of the plurality of pixels 3101 and horizontal scan circuits 3105, 3110 for forwarding the stored pixel signals in response to horizontal shift pulses input from input terminals 3122, 3123. In addition, the readout channels 3071, 3072 also have amplifiers 3107, 3112 for amplifying the signals that are read out and clamps 3124, 3125 for clamping the amplified signals at a particular electric potential.
A description is now given of the operation of the conventional fixed image sensing apparatus having the structure described above, with reference to FIG. 11.
First, when light strikes each of the plurality of pixels 3101, the pixels 3101 generate pixel signals of a level determined by the amount of incoming light. Next, pixel signals read out from odd-numbered columns in a row of pixels 3101 selected by the vertical scan circuit 102 are stored in the line memory circuit 3104, and at the same time pixel signals read from even-numbered columns in the same row are stored in the line memory circuit 3109.
Next, the horizontal scan circuit 3105 inputs a horizontal shift pulse from either outside or inside the chip from the input terminal 3122. Based on the input horizontal shift pulse, the pixel signals read out to the line memory circuit 3104 are then sequentially selected and output to the amplifier 3107. At the amplifier 3107, the input pixel signals are amplified and output to a processing circuit (not shown in the diagram) from an output terminal 3108.
Similarly, the horizontal scan circuit 3110, based on a horizontal shift pulse input from the input terminal 3123, sequentially selects pixel signals read out to the line memory circuit 3109 and outputs them to the amplifier 3112. At the amplifier 3112, the input pixel signals are amplified and output to a processing circuit (not shown in the diagram) from the output terminal 3113.
In addition, the dark level signals output from the optical black pixels within the plurality of-pixels 3101 are then clamped at a desired electric potential using clamps 3124, 3125. Further, at each of output terminals 3108 and 3113, switches 3116 and 3117 connected in parallel are switched ON/OFF in alternating sequence so as to output pixel signals from the odd-numbered columns of pixels and the even-numbered columns of pixels from the output terminal 3120 via the output buffer circuit 3119.
FIG. 12 shows horizontal shift pulses 1 and 2 input at input terminals 3122, 3123 of FIG. 11, a dark level signal and a pixel signal output from output terminals 3108 and 3113, a clamp pulse clamping, the ON/OFF action of the switches 3116 and 3117, and a dark level signal and a pixel signal output from output terminal 3120. FIG. 12 shows a state in which a pulse wave is input to input terminals 3122, 3123 in, for example, 6 clock parts each.
In FIG. 12, of the signals output from output terminal 3120, the pixel signals and dark level signals read out from pixels of a given row of columns 1-12 are assigned reference numerals (1)-(12), respectively. Also, pixel signals of output terminals 3108 and 3113 are given reference numerals corresponding to those of the signals at output terminal 3120. Reference numerals (1)-(6) correspond to dark level signals obtained from the optical black pixels and reference numerals (7)-(12) correspond to the pixel signals from the effective pixels.
According to FIG. 12, the dark level signals (1), (3) and (5) and the pixel signals (7), (9) and (11), synchronized to horizontal shift pulse 1, are sequentially output at the output terminal 3108. Similarly, dark level signals (2), (4) and (6) and pixel signals (8), (10) and (12), synchronized to horizontal shift pulse 2, are sequentially output at the output terminal 3113. By activating the clamps 3124, 3125 at the point at which the dark level signals (1) and (2) are output from the output terminals 3108, 3113, the dark level signals are clamped at a desired electric potential.
Next, by switching switches 3116 and 3117 ON/OFF in alternate succession, the dark level signals (1)-(6) and the pixel signals (7)-(12) are output at the output terminal 3120. By so doing, although the output terminals 3108 and 3113 operate at half-cycle with respect to the clock rate at output terminal 3120, the readout speed can be increased relatively easy.
Moreover, when reading out signals using multiple channels as described above, a structure that always reads out signals of the same color from the channels is disclosed in Japanese Patent Application Laid-Open No. 9-46480, and a method for correcting offset error at each channel is disclosed in Japanese Patent Application Laid-Open No. 2001-245221.
Moreover, Patent Application Japanese Laid-Open No. 5-328224 discloses a structure using multiple channels to read a plurality of pixels in the horizontal direction and using a switch to perform time division multiplexing on the signals read by the multiple channels. According to such a structure, even if the readout speed at each of the channels is slow, the charge readout can be read at high speed and the number of terminals can be reduced by time division multiplexing of the read-out electric charge signals.
However, Japanese Laid-Open Patent Application No. 9-46480 and 2001-245221 have a disadvantage in that they increase the number of output pins because four or five output pins are required for each output terminal. In addition, Japanese Laid-Open Patent Application No. 5-328224 has the following problem, described with reference to FIG. 13.
FIG. 13 is a diagram of the structure of another conventional image sensing apparatus, illustrating the adaptation of the structure disclosed in Japanese Laid-Open Patent Application No. 5-328224 to a color readout. For the sake of simplicity, two horizontal scan circuits are used to read out a charge from two pixels at a time. In FIG. 13, a plurality of pixels 1 are covered by a Bayer arrangement filter, with the G-B pixel columns being read by the first horizontal scan circuit 3 and the R-G pixel columns-being read by the second horizontal scan circuit 4.
In a case in which, as depicted here, an Nth line is selected and read out by the vertical scan circuit 2, a G signal of every other pixel is continuously output from a first differential amplifier 5 and an R signal of every other pixel is continuously read out from a second differential amplifier 6. When time division multiplexed by a multiplexer 7, these G and R signals are output in alternating sequence from an output terminal (OUT).
However, by outputting signals of different colors from a single output terminal using such multiplexing as described above, there is a risk that the two colors will mix, and in any case such an arrangement complicates downstream signal processing circuit structures for operations such as signal separation outside semiconductor image sensing apparatuses.
The reason is as follows: Parasitic resistance R and parasitic capacitance C occurs in the wires inside the image sensing apparatus, and a change in electric potential in such wiring can be explained as a transient phenomenon. That is, the electric potential change in wiring with such parasitic elements is determined by the parasitic resistance R and the parasitic capacitance C, and with a time constant CR, a V(t) can be expressed by equation (1):V(t)=Voε−(1/RC)t  (1)where Vo is the electric potential in a steady state of the wiring and ε is a natural constant.
As can be understood from equation (1), V(t) changes exponentially with time and approaches Vo.
Thus, the waveform output from the output terminal 3120 of FIG. 11 (output terminal 3120 of FIG. 12) has a different output level at (7) and (8), so it takes time for the electric potential to fall from (7) to (8), as is the case with the output level in the transition from (6) to (7). One of the reasons for the large differences in the continuous output level at the output terminal 3120 is that the outputs from terminals 3113 and 3108 are outputs from pixels of color filters of different transmissivity.